GAISE + SBI:
Experiencing Statistics Instruction
from the Student Perspective
Friday, July 31, 2026 · 3:00–5:00 pm
Jen McNally · Laura Callis · Karen McGaughey
INSTITUTE FOR INTRODUCTORY STATISTICS INSTRUCTORS�SPONSORED BY EAPOST: EXPANDING THE ART & PRACTICE OF STATISTICAL THINKING
Session Roadmap
2
1
3:00–3:20
Welcome & GAISE Framing
2
3:20–3:45
SBI Activity — as Students
3
3:45–4:15
Debrief Part 1 — The Statistics
4
4:15–4:50
Debrief Part 2 — The Pedagogy
5
4:50–5:00
Bridge to Saturday
The Central Question
What does it look like —
and feel like —
when a statistics course is
designed around student
reasoning and genuine
uncertainty?
3:00–3:20
Think about a statistics class…
4
Think of a statistics class you took — or a lesson you've taught or observed.
What made students actually engage with the material?
What made them tune out?
60 seconds — think on your own, then 3–4 share out
3:00–3:20
The GAISE College Report
5
Guidelines for Assessment and Instruction in Statistics Education
1
Teach statistical
thinking
2
Focus on conceptual
understanding
3
Integrate real data
with context & purpose
4
Foster
active learning
5
Use technology to
explore concepts
6
Use assessments to
improve learning
Keep these in mind — you'll be identifying them in your own experience shortly.
3:20–3:45
You are now a student.
SBI Activity — Experiencing Simulation-Based Inference
SBI Activity
The Research Question
7
Can people correctly identify AI-generated text more than half the time?
Researchers claim that human readers can detect AI-generated writing at a rate better than chance.
We're going to test that claim — using ourselves as the subjects.
SBI Activity — Phase 1
Read & Classify — on your own
8
1
Read
each of the 6 text samples on your handout
2
Judge
each as AI-generated or Human-written — circle your answer
3
Count
how many you got correct (answer key revealed after)
4
Calculate
the proportion you got correct. Who did the best in your group?
No collaboration during this phase.
SBI Activity — Phase 1
Who did the best? How many correct?
8
Do you think this is evidence that this person can detect AI created content? Why or why not?
No collaboration during this phase.
SBI Activity — Phase 2
Setting Up the Simulation
10
Null Hypothesis H₀
p = 0.5
Humans are just guessing — detection is no better than chance.
Alternative Hypothesis Hₐ
p > 0.5
Humans can detect AI text at a rate better than chance.
Open: rossmanchance.com/applets/2021/oneprop/OneProp.htm
SBI Activity — Phase 3
Simulation Set Up
11
Why flip a coin?
Number of tosses
6
One flip = one question
What does heads mean?
Put your results on the sticky note and post on the board
Run 1000 simulations → examine the null distribution → shade the tail → read the p-value
So far, do we think that our colleague could have gotten these results just by chance?
SBI Activity — Phase 3
Applet Settings
11
Heads
0.50
null hypothesis
Number of tosses
6
6 flips = 1 test
Number of repetitions
1000
simulated null world
Shade direction
Right tail
one-sided alternative p > 0.5
Run 1000 simulations → examine the null distribution → shade the tail → read the p-value
SBI Activity — Phase 4
Interpret & Conclude
12
Q1
What does the p-value tell us here? What does it not tell us?
Q2
What population can we generalize to? What type of tasks can we generalize to?
3:45–4:15
Debrief Part 1
Unpacking the Statistics
Debrief 1
The 3S Strategy
14
The inferential framework you just used — now named.
S
Statistic
Summarize the data
Proportion correct
(e.g., p̂ = 5/6 = 0.83)
S
Simulate
Assume the null; generate
a distribution
1000 simulations with p = 0.5
in the Rossman-Chance applet
S
Strength of evidence
Compare statistic to
null distribution
How far is 0.83 in the tail?
What's the p-value?
Debrief 1
Discussion: Statistical Reasoning
15
Highest priority questions marked ★
Q1
In what ways did we engage in the statistical process?
Q2
What would change if we had 10 questions instead of 6?
Q3
What would change if the questions had 2 distractors (wrong answer choices)?
Q4
Suppose someone got 0 out of 6 right. This is unlikely to happen due to chance. How would you think differently about the investigation?
Q5
What assumptions are baked in? Are they reasonable here?
4:15–4:50
Debrief Part 2
The Pedagogical Lens
Debrief 2
Step Back: The Instructor View
17
Turn to a partner. Take 3 minutes.
Answer: What did that lesson ask you to do, as a learner?
Not just what you learned — what were you asked to DO?
Debrief 2
GAISE — Where Did You See It?
18
Build this together — then show the table.
1. Statistical thinking
Reasoned about uncertainty — interpreting what the result means (and doesn't)
2. Conceptual understanding
Null distribution built visually; 3S scaffolds the logic without shortcuts
3. Real data with context
We used our own responses; AI detection is a live 2026 question
4. Active learning
Individual → pair → class dotplot → simulation: no passive reception
5. Technology
Applet makes 1000 simulations visible; focus stays on interpretation
6. Assessment
Discussion questions probe understanding; misconceptions surface in conversation
Debrief 2
What Makes This Hard to Teach?
19
Most of us learned statistics through formulas and procedures. Our instinct when we teach is to show, then practice.
GAISE-aligned instruction often asks us to do the reverse, engage students in the statistical inquiry process.
Where does that feel uncomfortable for you as an instructor?
Q1
What would a student who learned statistics procedurally struggle with in a lesson like this?
Q2
What do you need to know — about simulation, the applet, inference — to teach this confidently?
Q3
Where in your own course could a lesson structured like this fit?
Extension
How can we prepare?
19
Q1
Look through the videos of students talking about the AI prompt.
Q2
What do you notice about their thinking?
Q3
What do you think you could do to capitalize on their rich thinking or adjust problematic thinking?
4:50–5:00
Tomorrow, you'll be in the room
as instructors, not students.
But keep tonight in mind.
Your students are sitting where you were sitting an hour ago.
What do you want them to walk out with?
Dinner — Marketplace · 5:00 pm